In the AI era, high-performing organizations need Toss-style TPMs

After establishing a TPM position together with the Tech Excellence Team in December 2024, Toss redefined the role as one that handles problems that can't be solved by schedule, risk, and dependency management alone.

As AI goes deeper into an organization, model adoption, data governance, quality standards, security, developer productivity, and ways of working all become entangled at once. As a result, gray-zone problems between teams and outside the boundaries of responsibility increase more rapidly, and the more an organization has high autonomy and high execution capability, the more it needs a role dedicated to covering these gaps.

While a traditional TPM is closer to delivering an already-defined program, a Toss-style TPM is someone who finds and redefines important problems that haven't yet been named, and structures them. They turn things into an execution structure down to which team should move, who should be the DRI, and what decisions and authority are needed.

• Find the problem first.
• Turn strategy into an execution structure.
• Remove blockers between teams.
• Look at people and structure together.
• Prove results through real-world change, not documents.

It emphasizes that becoming a good TPM requires the ability to structure ambiguous phenomena, the power to turn strategy into execution, influence to move people without formal authority, systems thinking to solve recurring problems through structure, and completeness to deliver results to the end. It makes clear that coordination is a necessary skill, but the identity lies in problem-solving.